Multiplex hybridization of a number of nucleic acid single strands resulting in the formation of many sequence specific, individual duplex complexes, is the central process in a number of nucleic acid based diagnostic assays. Such assays are currently employed to analyze gene expression, detect genetic variations and mutations, assess viral loads in response to therapeutic agents and detect infectious pathogenic agents. Ideally, each strand present in a multiplex reaction is meant to form a duplex with only its perfectly matched and complementary single strand. In reality however, depending on the particular sequences present, many of the resident single strands can also anneal with strands other than their perfectly matched complement strand, resulting in cross-hybridization, x-hyb. X-hyb is a major nemesis of multiplex hybridization reactions causing false positive signals, lowered accuracy and sensitivity. Although x-hyb is readily acknowledged as a potential problem by all practicitioners of multiplex hybridization assay, the molecular interactions responsible for x-hyb are not understood. Other than redesigning the probes and primers and trying them again (very much a trial and error, empirical approach) there is little recourse should a given probe and/or primer set fail to function as designed [due to x-hyb]. The goal of this Phase II project is to quantitatively define sequence dependent features of x-hyb and develop a robust analytical framework for diagnosing and predicting x-hyb on the level of individual sequences. The first specific aim involves evaluation of thermodynamic parameters for tandem mismatches as a function of sequence, in solution and on microarrays, then incorporating this information in a software toolset for x-hyb analysis. The second specific aim is to apply the capabilities gained to develop RT-PCR and microarray based assays for detection of an alternatly spliced variant of the Creld1 gene, the (exon 9b) isoform, whose presence has been has been linked to various cancers.